In the era of cloud computing, massive applications and services all are hosted on a cloud platform. In addition to ensuring high availability of the cloud platform itself, the cloud platform still needs to provide assurance of high availability for the applications and services hosted on the cloud platform. In an existing solution, high availability of an application or a service is usually implemented by deploying two primary instance nodes or by deploying a primary instance node and a secondary instance node. For a solution implementing high availability by using two primary instance nodes, instances of multiple applications are parallel, and all instance nodes can normally receive external request information and process the external request information. If an instance node of the instance nodes is faulty, load born by the instance node is shared by another instance node. A solution in which a primary instance node and a secondary instance node are used is a highly reliable solution that is most dominant in the industry. An application instance of a primary instance node is backed up on a secondary instance node. When the primary instance node is normal, all external requests are used for accessing the primary instance node, and the external requests are switched to the secondary instance node only when the primary instance node is faulty and cannot run normally.
It can be seen from the foregoing that, in the solution in which a primary instance node and a secondary instance node are used, the secondary instance node occupies a resource that is absolutely the same as a resource occupied by the primary instance node, but is in a standby state all the time, which causes a waste of physical resources. However, in the solution in which two primary instance nodes are used, a resource utilization rate of the two primary instance nodes cannot reach 50%. If the resource utilization rate reaches 50%, when a single point is faulty, all pressure is concentrated on the other active primary instance node, and excessively high pressure causes that a service cannot run normally. As a result, more usage wastes of node resources are caused.